Do nest boxes in restored woodlands promote the ......1 Restoration Ecology Do nest boxes in restored woodlands promote the conservation of hollow-dependent fauna? David Lindenmayer1,2,3,

1

Restoration Ecology

Do nest boxes in restored woodlands promote the conservation of hollow-dependent

fauna?

David Lindenmayer1,2,3, Mason Crane1, Wade Blanchard1, Sachiko Okada1 and Rebecca

Montague-Drake1

1Fenner School of Environment and Society, The Australian National University, Canberra,

ACT 2601, Australia

2ARC Centre of Excellence for Environmental Decisions, The Australian National

University, Canberra, ACT 2601, Australia

3National Environmental Science Program Threatened Species Recovery Hub, The Australian

National University, Canberra, ACT 2601, Australia

Corresponding author: David Lindenmayer. Email: [email protected]

Author contributions: MC, DBL designed the study; MC, SO, RMD completed field work;

WB in collaboration with DBL, MC conducted the data analysis; DBL, MC, WB wrote the

manuscript.

Running title: Nest box use in woodlands

mailto:[email protected]

2

Abstract

Vegetation restoration is considered an important strategy for reversing biodiversity decline

in agricultural areas. However, revegetated areas often lack key vegetation attributes like

large old hollow-bearing trees. As these trees take a long time to develop, artificial cavities

such as nest boxes are sometimes provided to address lag effects. We conducted a 3-year

experiment using 150 nest boxes with four designs to quantify patterns of occupancy within

16 replanted areas and 14 patches of remnant old growth eucalypt woodland. We quantified

patterns of occupancy of nest boxes in physically connected versus isolated remnants and

plantings, and multiple covariate effects on nest box occupancy at the nest box, tree, patch

and landscape levels. Our analyses revealed a lower probability of nest box occupancy within

remnants (versus plantings) for two of the six response variables examined: any species, and

the Feral Honeybee. Nest boxes in connected remnants and plantings were more likely to be

occupied than those in isolated plantings and remnants by any mammal and the Common

Brushtail Possum. Nest boxes in restored woodlands are used by some hollow-dependent

fauna, but principally already common species, and not taxa of conservation concern. Nest

boxes also were used by pest species. A key management consideration must be to create

connected habitat to facilitate colonization of nest boxes by mammals. Approximately 15%

of the cavity-dependent vertebrates within the study area used next boxes, possibly because

the diverse requirements of the array of other species were not met by the range of nest boxes

deployed.

Keywords: Cavity-users, connectivity, hollow-dependent animals, large old trees, vegetation

restoration, agricultural landscapes

3

Implications for Practice

• Restored areas often lack key attributes that are critical for biodiversity. Large old

trees with hollows are one of these key attributes.

• The establishment of nest boxes within revegetated areas is one potential practical

strategy to promote colonization by cavity-dependent wildlife.

• The connectedness of sites influenced nest box occupancy and appears to be

important for improving the effectiveness of nest box programs, particularly for some

species of arboreal marsupials.

• Nest boxes primarily benefitted already common species or pest species. Taxa of

conservation concern may require highly targeted species-specific nest box designs

and/or prolonged periods of time to colonize nest boxes.

Introduction

Millions of hectares of land worldwide are in need of restoration (Clewell & Aronson 2007;

Minnemeyer et al. 2011; Menz et al. 2013; Suding et al. 2015), particularly in agricultural

areas where extensive native vegetation clearing has led to a wide range of environmental

problems including land degradation and biodiversity loss (Karp et al. 2012; Loos et al. 2014;

Latawiec et al. 2015). Vegetation restoration is considered to be an important strategy for

reversing biodiversity decline in agricultural areas (e.g. Bullock et al. 2011; Cristescu et al.

2012). However, the effectiveness of restoration for biodiversity still needs to be carefully

quantified (e.g. Ray Benayas et al. 2009; Catterall et al. 2012; Wortley et al. 2013). Indeed,

revegetated areas often lack key attributes of vegetation structure like large old hollow-

bearing trees that take a long time to develop (Vesk et al. 2008) and which provide crucial

habitat structures that biota depend on for survival (e.g. hollows, fallen woody debris, and

decorticating bark microhabitat) (Gibbons et al. 2008; Fischer et al. 2010; Crane et al. 2014).

4

A paucity of these key resources may mean that a significant proportion of the fauna that

might otherwise inhabit areas of natural vegetation in agricultural areas will be absent

(Flaquer et al. 2006; Cunningham et al. 2007). The provision of artificial cavities such as nest

boxes is one widely employed approach that attempts to address this problem of lag effects in

the time needed to recruit large old hollow-bearing trees (Beyer & Goldingay 2006;

Goldingay & Stevens 2009) including in restored areas in agricultural landscapes (Goldingay

et al. 2015). To date there is limited information on the effectiveness of nest boxes in

recovering biodiversity in restored areas.

In this study, we quantified patterns of occupancy of nest boxes within replanted areas

and compared them with matched patches of remnant old growth temperate eucalypt

woodland (sensu Lindenmayer et al. 2012). We focused our study on the temperate woodland

biome of the South West Slopes of New South Wales, south-eastern Australia. We posed the

key question: Are there differences in nest box occupancy between woodland remnants and

plantings? At the outset of this study, we postulated that rates of occupancy would be

significantly higher in nest boxes established within plantings than in remnants. This was

because previous studies in other vegetation types such as forests and plantations (see Smith

& Agnew 2002; Lindenmayer et al. 2009) have found that hollow-dependent animals are less

likely to use nest boxes when natural cavities are more readily available (as occurs in this

study within woodland remnants that are dominated by large old trees).

Vegetation cover in many agricultural landscapes (including in our study area) has been

extensively cleared and fragmented (Gibbons & Boak 2002). As a result, areas of both

remnant native woodland and replantings are often physically disconnected from other areas

of native vegetation. This may, in turn, affect movement and hence patch occupancy patterns

by a range of fauna, including hollow-dependent taxa that might otherwise potentially use

nest boxes (Cooper et al. 2002; van der Ree et al. 2004; Doerr et al. 2010). On this basis, a

5

key additional question in our investigation was: Are there differences in the occupancy of

nest boxes among remnants and plantings that are physically connected to other areas of

native vegetation versus those which are isolated? At the outset of this investigation, we

postulated that such differences in physical connectivity (sensu Lindenmayer & Fischer

2006) would influence nest box occupancy for dispersal-limited species such as arboreal

marsupials, but not for more mobile taxa like the majority of birds and invertebrates such as

the Feral Honeybee (Apis mellifera).

We also sought to determine if there was an interaction between broad vegetation type

and connectedness effects. That is: Are there differences in occupancy rates of nest boxes

between connected and unconnected plantings versus those in connected and unconnected

remnants? If both design variables (viz: broad vegetation type and connectedness) are

important, then the highest rates of nest box occupancy would be predicted to occur in

connected plantings and the lowest in unconnected remnants.

In addition to addressing the three questions outlined above, we also quantified the

effects of other covariates at box, site and landscape-level. These included the entrance type

and other physical characteristics of nest boxes, density of stems at a site, distance of a site

(i.e. a remnant or planting) to a watercourse, and the number of large old scattered paddock

trees in the landscape surrounding a given site. A paddock tree was defined as a scattered tree

in an otherwise cleared or semi-cleared agricultural field (or paddock) (sensu Manning et al.

2006). We explored the effects of these covariates as they have been found to be important in

other studies of nest boxes (e.g. Finch 1989; Fargallo et al. 2001; Smith & Agnew 2002;

Durant et al. 2009; Goldingay & Stevens 2009; Goldingay et al. 2015).

Nest boxes are a widely recommended management activity for restored areas of

temperate woodland in many parts of Australia. Our hope is that the new information

6

presented in this paper will assist closing key knowledge gaps associated with the twin goals

of vegetation and wildlife restoration in Australian agricultural landscapes.

Methods

Study area and study design

We conducted this study in the Junee district of southern New South Wales, south-eastern

Australia (Figure 1). The district is highly modified for agriculture and the majority of the

former cover of native vegetation has been cleared to make way for dryland cropping and for

grazing livestock. The remaining native vegetation occurs predominately along roadsides,

within riparian zones, as small patches of paddock trees or as scattered paddock trees (Crane

et al. 2014). Over the last 30 years farmers have been attempting to address the lack of native

vegetation by establishing native vegetation plantings.

Our study encompassed 150 nest boxes located on 30 sites each with 5 nest boxes of different

designs. The 30 sites comprised seven connected plantings, nine isolated plantings, eight

connected remnants and six isolated remnants. We classified sites as isolated if there was a

gap > 70m to an area of native vegetation. This value was based on previous studies that have

indicated that gaps in native vegetation can significantly impede movement of animals such

as arboreal marsupials. This is because gliding marsupials are unable to volplane between

widely spaced trees (van der Ree et al. 2004) – especially in woodlands and plantings where

tree height is limited to 30 m (and often much shorter) which limits gliding distance (as it is a

function of tree height; Lindenmayer 2002).

The plantings in our study were typically 15-25 years old, with tree heights 12-15 m

tall. Plantings were characterized by a mix of locally endemic and exotic Australian ground

cover, understorey and overstorey plant species (primarily Eucalyptus and Acacia spp). Most

plants were typically spaced 2 m apart, but there was not a standard set of spacing and plant

species composition protocols applied in revegetation efforts. There was an average of 0.15

7

hollow-bearing trees per ha in the plantings. The remnant patches in our study were

dominated by Box-gum woodlands and occurred along roadsides and as small patches of

trees in an otherwise highly modified cropping or grazing paddocks. There was an average of

2.11 hollow-bearing trees per ha in the remnants.

We erected nest boxes in February 2010. At each site, we attached nest boxes to living

and dead trees between 3 and 6 metres above the ground. We deployed four different box

designs (see Table S1) that were based on designs previously used to accommodate particular

species, the Common Brushtail Possum (Trichosurus vulpecula), Squirrel Glider (Petaurus

norfolcensis), the Superb Parrot (Polytelis swainsonii) and the Laughing Kookaburra (Dacelo

novaeguineae). We modified the glider and kookaburra boxes by adding a 30mm cavity in

the back wall as an experimental bat chamber. The nest boxes were constructed from marine

plywood. We installed nest boxes within 200m of each other and at each site supported one

Common Brushtail Possum (BP) box, two Squirrel Glider (SG) boxes, one Superb Parrot

(SP) box and one Laughing Kookaburra (KB) box or two BP boxes, two SG boxes, and one

SP box.

We checked nest boxes on four occasions: October (spring) 2010, December/January

(summer) 2010/11, October (spring) 2011 and December/January (summer) 2012/13. These

periods corresponded to times when many cavity-dependent animals are breeding and there is

a high chance of detecting them. We determined usage from the presence of an animal, scats,

hairs, feathers, nest, eggs or a combination of methods. In the absence of an animal, species

identify was determined through scat or hair analysis by an expert (Barbara Triggs) who

assigned a level of confidence to each record (definite, probable or uncertain). We restricted

our analyses to data on animals that were physically observed and scat and fur samples

deemed to be “definite”.

8

We measured finescale covariates in the field and calculated broadscale covariates from

spatial data layers in a GIS for subsequent use in modeling of the factors influencing nest box

occupancy. We explored these two scales of variables because choices by land managers

about locating nest boxes can often be made at both a broadscale (e.g. which farms) and at

finescale (e.g. which patch and which tree within patch).

Finescale variables were attributes of a given nest box, tree (as an attachment site) or

site within which nest boxes where established and they included the diameter of tree on

which a nest box was attached, the level of dieback of the tree on which the nest box was

attached, the total number of stems at a site, the number of trees greater than 50 cm in

diameter at a site, number of hollow bearing trees at a site and a lithology fertility rating.

Broadscale variables characterized the landscape surrounding locations where nest

boxes were established and they included the number of paddock trees within 500 m of the

site, the distance from a site to a drainage line, topographic wetness index (TWI), and the

distance to the closest major patch of native vegetation. The topographic wetness index is a

continuous terrain-based measure of likely moisture contributed to a site as a result of an

area’s position in the landscape, ranging from negative values on ridges (with no contributing

catchment) and upper slopes (small contributing catchment/steep slope) to increasingly

higher positive values through lower slopes, valley flats and eventually drainage lines.

Statistical analyses

We grouped the species recorded in the nest boxes into five broad overlapping categories.

Specifically, we analyzed the presence in the next boxes of the following groups: marsupials

(Antechinus, Common Brushtail Possum, Common Ringtail Possum, Sugar Glider),

mammals (marsupials plus the Lesser Long-eared Bat and the exotic Black Rat), birds

(Cockatiel, Common Starling, Eastern Rosella and Galah), other species – non-mammal or

bird (Feral Honeybees, Peron’s Tree Frog and Marbled Gecko) and any species detected. We

9

also modeled the three individual species with sufficient presence data to warrant further

individual analysis (Common Brushtail Possum, Common Starling and Feral Honeybees).

We modeled the effects of four broadscale or finescale (site-level) variables and one

interaction: survey occasion (spring 2010, spring 2011, summer 2011 and spring 2012);

connectivity (connected and isolated); vegetation type (planting versus remnant); Number of

paddock trees within 500 meters; and the interaction between connectivity and vegetation

type. We also modeled the effects of nest box type (BP, SG, SP, KB), tree diameter, dieback

score, log of the total number of stems, number of trees greater than 50 cm and within 50 m,

number of hollow bearing trees within 50 m, distance to drainage line, topographic wetness

index (TWI), lithology fertility rating, distance to closest major vegetation. The response

variable for all analyses was the presence/absence of the species or species group of interest

which we modeled using a binary logistic regression with a random effect for site. We used

the glmer function from the lme4 package (Bates et al. 2014) to model the presence/absence

of both the individual species and groups.

We used Akaike Information Criterion (AIC) to guide model selection on the logistic

regression. We chose AIC over the Bayesian Information Criterion (BIC), at this preliminary

stage, to allow the inclusion of more potential predictors in the model. Due to the more

stringent inclusion criteria with larger sample sizes, BIC tends to favor simple models

compared to AIC.

Due to the large number of potential predictor variables (14 plus an interaction), we

employed the following two-part variable selection strategy. We used the package MuMIn

(Barton 2014) to explore all possible subsets of the site level variables. We then retained the

variables from the best fitting AIC model and carried them to the second stage of model

selection. In the second stage, we then fitted all possible models from the next box-level

variables while keeping the variables from the site-level stage in each of the models.

10

The final models were then fitted using the package MCMCglmm (Hadfield 2010). The

MCMCglmm package fits the logistic regression model via Markov Chain Monte Carlo

(MCMC) techniques and gives samples from the posterior distribution. We chose

uninformative but proper priors for the fixed effects components and minimally informative

but proper priors for the variance components. Specifically, we used multivariate normal

priors for the regression parameters and inverse Wishart distributions for the variance

components.

The logistic regression model parameters are summarized by the posterior mean, 95%

credible intervals and Btail, which gives the fraction of the posterior distribution that is to the

left(right) of zero if the posterior mean is greater(less) than zero. Small values of Btail

indicate support for non-zero parameter values, that is, posterior distributions that are shifted

away from zero. We report the parameters from the presence and conditional abundance

components of the hurdle more on the log odds ratio and log scale respectively.

We also assessed the residuals from of the logistic regression models for evidence of

nonlinearities over and above specified by our models using generalized additive models

(Wood 2006). In all cases there was no evidence of non-linearities.

Results

General findings

We recorded a high level of usage of the 150 nest boxes over the three years of our

investigation (Table 1). We recorded 13 species of animals using nest boxes, including seven

six species of native mammals: the Yellow-footed Antechinus (Antechinus flavipes; 2

detections), Sugar Glider (Petaurus breviceps; 2 detections), Common Brushtail Possum

(Trichosurus vulpecula; 52 detections), Common Ringtail Possum (Pseudocheirus

peregrinus; 8 detections), and Lesser long-eared Bat (Nyctophilus geoffroyi; 4 detections),

and one introduced species – the Black Rat (Rattus rattus; 24 detections). The four bird

11

species detected using nest boxes were the Galah (Eolophus roseicapillus; 1 detection),

Cockatiel (Nymphicus hollandicus; 1 detection), Eastern Rosella (Platycercus eximius; 23

detections) and the exotic European Starling (Sturnus vulgaris; 116 detections). The

remaining two species of vertebrates detected were the Marbled Gecko (Christinus

marmoratus; 2 detections) and Peron’s Tree Frog (Litoria peronii; 6 detections). The Feral

Honeybee (Apis mellifera; 71 detections) was the sole species of invertebrate that was

identified to species level in this study.

Key response variables influencing the occupancy of nest boxes

We completed detailed statistical analyses of design variables and nest box, site and

landscape-level covariates influencing six response variables; the occurrence of the Common

Brushtail Possum, the presence of any mammal species, the occurrence of the exotic

Common Starling, the presence of any bird species, the occurrence of the exotic Honeybee,

and the presence of any species. Models showing all effects are summarized in Appendices 1

and 2.

Broad vegetation type differences – plantings versus remnants

Our analyses revealed a lower probability of presence in a nest box within remnants (versus

plantings) for two of the six response variables: any species (Btail = 0.014), and the Feral

Honeybee (Btail < 0.001). The broad vegetation effect remained important only for the Feral

Honeybee (Btail = 0.047) after fine scale variables were included in the final model (Figure

2A, Appendix 2).

Connected versus unconnected plantings and remnants

Analyses of broad scale variables indicated that nest boxes in connected remnants and

plantings were more likely to be occupied than isolated plantings and remnants by any

mammal (Btail = 0.029) and the Common Brushtail Possum (Btail < 0.001). The reverse

effect was observed for the Feral Honeybee (Btail =0.086). These effects remained

12

unchanged after fine scale variables were included in the final model (Figure 2B).

Connectedness was not important in models based on either broad scale variables or the final

models that included fine-scale variables for any bird response variable or for the Common

Starling (Figure 2B).

We found no evidence of interaction effects between broad vegetation type and

connectedness for any of the response variables we analysed.

Other effects

We found that nest box characteristics had an important effect on occupancy for the majority

of response variables examined. The lowest rates of occupancy were in KB nest boxes for the

Feral Honeybee (KB vs BP Btail = 0.018, KB vs SG Btail = 0.006), Common Starling (KB vs

BP Btail <0.001, KB vs SG Btail <0.001), the presence of any bird species (KB vs BP Btail

<0.001, KB vs SG Btail <0.001), and the presence of any species (KB vs BP Btail <0.001,

KB vs SG Btail <0001). The lowest rates of occupancy for the Common Brushtail Possum

(SG vs BP Btail <0.001, SG vs KB Btail <0.001) and the presence of any mammal species

(SG vs BP Btail <0.001, SG vs KB Btail = 0.004) were in SG nest boxes (Appendix 2).

Our analyses revealed that survey year effects were prominent in the final models for

almost all of the responses variables we examined. The lowest probability of occurrence of

the three species we analyzed (Common Brushtail Possum, Common Starling and Feral

Honey Bees) and the three composite measures (any mammal, any bird, and any species) all

were lowest in the first year of survey (summer 2010). The Common Brushtail Possum and

any mammal experienced peak nest box occupancy in spring survey of 2011. By contrast, the

greatest occupancy rate for the Common Starling, the Feral Honeybee and any bird species

was in summer 2012 (Table S1, Appendices 1 and 2).

We found that the presence of any mammal (Btail < 0.001), and the Common Brushtail

Possum (Btail = 0.002) were negatively associated with distance to a watercourse (Appendix

13

1; Figure 2C). Other variables featured in final models included: (a) A negative effect of

distance to major block of native vegetation for any species (Btail =0.002) and the Feral

Honeybee (Btail = 0.070); (b) A positive association with the dieback score and the presence

of any mammal (Btail = 0.026), and the presence of the Common Brushtail Possum (Btail =

0.014); (c) A positive association between the number of stems at a site and the presence of

the Common Brushtail Possum (Btail = 0.28) and the presence of the Feral Honeybee (Btail =

0.034), and; (d) A negative association between the number of paddock trees and the

presence of the Feral Honeybee (Btail = 0.066). Models showing these various effects are

summarized in Appendix 2.

Discussion

Large areas of highly modified agricultural land worldwide have been targeted for vegetation

restoration as part of attempts to tackle problems such as land degradation and biodiversity

loss (Ray Benayas et al. 2009; Lamb 2011; Menz et al. 2013). This is true in large parts of

southern Australia where such problems are widely recognized (Hajkowicz 2009; Munro &

Lindenmayer 2011). Time lags in the development of key structural attributes of the

vegetation in restored areas potentially limits their value for some groups of animals such as

hollow-dependent vertebrates (Cunningham et al. 2007; Vesk et al. 2008). In an attempt to

counter this problem, the establishment of nest boxes within revegetated areas is a widely

recommended management action in many parts of Australia (Durant et al. 2009; Goldingay

et al. 2015). However, the effectiveness of nest box establishment in promoting biodiversity

conservation within restored woodlands is poorly known, in part because the factors affecting

occupancy and use have often not been documented in designed and implemented studies.

We addressed three key questions as part of this investigation. The answer to our first

question: Are there differences in nest box occupancy between woodland remnants and

plantings? – was generally no. Broad vegetation type effects were found for only two of the

14

six response variables we examined and then remained important only for the Feral

Honeybee after fine scale variables were included in the final model. This result was

unexpected as we postulated that rates of occupancy would be significantly higher in

plantings than in remnants because the former broad vegetation type support fewer hollow-

bearing trees. The reason for the paucity of broad vegetation effects remain unclear. It is

possibly related to the fact that the woodlands in our study have been heavily altered and

support significantly depleted numbers of hollow-bearing trees relative to unmodified

woodlands (Gibbons et al. 2010). These woodland areas typically support fewer hollow-

bearing trees per unit area than forests where most previous studies have been conducted and

which show inverse relationships between nest box occupancy and the abundance of hollow

trees (e.g. see Lindenmayer et al. 2009). Therefore, animals in woodland remnants (and

plantings) may simply occupy nest boxes as they encounter them (Menkhorst 1984), resulting

in a general lack of broad vegetation type differences as found in our study.

The second key question in our study was: Are there differences in the occupancy of

nest boxes among remnants and plantings that are physically connected to other areas of

native vegetation and those which are isolated? The answer to this question was that

connectedness was generally important for nest box occupancy by mammals (any mammal,

or the Common Brushtail Possum) but not for birds. This result was possibly associated with

differences in mobility between arboreal and scansorial mammals and birds. Other studies of

mammals have suggested that physical connections between areas of vegetation play an

important role in patch occupancy in semi-cleared agricultural landscapes (e.g. van der Ree &

Bennett 2003; van der Ree et al. 2004; Goldingay et al. 2013). Surprisingly, we identified a

negative impact of connectedness on nest box occupancy by the Feral Honeybee. However,

this effect disappeared once fine-scale variables had been incorporated in the final model,

15

suggesting that other factors associated with individual boxes (e.g. entrance size) and sites

(e.g. stem density) outweigh the effects of physical connectedness for this species.

Our third question related to potential interaction effects between broad vegetation type

and connectedness effects on nest box occupancy. No such effects were identified for any of

the array of response variables subject to detailed analysis. To some extent, this result was

unsurprising given that main effects for broad vegetation type were rare and connectedness

effects were primarily confined to responses for mammals (see above).

Several tree and site-level covariates were important for some species and species

groups. The use of nest boxes by the Common Brushtail Possum and mammals in general

were significantly higher in sites closer to watercourses. This is likely the result of higher

species abundance and/or the provision of high quality habitat in the mesic parts of the

landscape, as has been shown for a number of arboreal and scansorial mammals (Soderquist

& MacNally 2000; Crane et al. 2012). The use of nest boxes by the Common Brushtail

Possum and mammals per se, also increased with elevated levels of ‘dieback’ in the tree to

which a given nest box was attached. It is not clear if this effect is driven by a preference for

trees of poor health or if it reflects some other (unmeasured) issues affecting tree health in

areas selected by these species.

Our study revealed that nest boxes were used by a range of hollow-dependent fauna.

However, the number of species which occupied boxes was ~15% of the total number of

cavity-dependent vertebrates (excluding bats) that repeated survey work over the past decade

has shown can occur in the temperate woodlands in the South West Slopes region, including

the Junee area where this investigation was completed. We also note that almost none of the

species recorded using nest boxes in our study were of conservation concern, in fact three of

the most frequently recorded taxa were exotic. Cavity-dependent species of conservation

concern such as the Superb Parrot, Brown Treecreeper, and Squirrel Glider were absent from

16

our surveys. This was despite two of the kinds of nest boxes deployed being specifically

constructed for two of these species (the Squirrel Glider and the Superb Parrot). We note that

while there are many records of the Superb Parrot and Brown Treecreeper from areas within

1-2 km of our study sites, there are none of the Squirrel Glider. Other researchers working

elsewhere have recorded a high frequency of use of nest boxes by the Squirrel Glider (e.g.

Beyer & Goldingay 2006; Goldingay et al. 2015). More tailored designs specifically to meet

the requirements of particular animals of conservation concern may be appropriate if a

management objective is to cater to the needs of animals of conservation concern. For

example, a more tailored design for the Squirrel Glider would be a nest box with a rear-entry

(Goldingay et al. 2015), although this would be of limited value in the particular area of our

study given its apparent absence from the region. Lag effects in the use of nest boxes may be

an additional or alternative explanation for the low rates of occupancy for some species of

conservation concern. Our data show that the lowest probability of occupancy was in the first

survey after establishment (2010) suggest that nest boxes may not have been discovered by

animals. Delayed occupancy has been observed in other nest box studies and a longer term

study in woodlands may be required to determine if greater rates of colonization by species of

conservation concern occur over time. Finally, even in the absence of species of conservation

concern, nest boxes can nevertheless be important for attracting other native animals like the

Sugar Glider and Yellow-footed Antechinus which play key ecosystem service roles such

insect pest control, pollination and are prey to large owls (Goldingay et al. 1991;

Lindenmayer 2002).

A key issue with the provision of nest boxes is the risk of creating additional nesting or

sheltering habitat for pest species (Pell & Tidemann 1997; Gibbons & Lindenmayer 2002),

(but see Goldingay et al. 2015). Our data suggested that this problem is a legitimate concern

in temperate woodland environments as three of the most commonly recorded individual

17

species were exotic species that are widely regarded as important pest animals – the Black

Rat, Common Starling, and the Feral Honeybee. We suggest that one approach to limit nest

box use by these species will be to ensure they have characteristics which make them

unsuitable for pest species (Goldingay et al. 2015).

In summary, our study has shown that nest boxes can support the occupancy of some

hollow-dependent species in plantings, but not at levels different to those observed in

remnants of temperate eucalypt woodland. The connectedness of sites targeted for nest box

establishment can have an important positive effect on the probability of occupancy and this

appears to be an important consideration for attempts to improve the effectiveness of nest box

programs. However, nest boxes in this study generally benefited already common species,

including a number of pest species. In contrast, species of conservation concern were

typically not recorded. A relatively small fraction of the overall total cavity-dependent fauna

in our study region occupied nest boxes. This may have occurred because a limited range of

nest box designs were employed, some plantings were not connected to other areas of native

vegetation, and the relatively limited period that nest boxes had been established.

Acknowledgments

We acknowledge New South Wales Roads and Maritime Services for financial support of

this project. The Murray Local Land Services and the Riverina Local Land Services also

provided indirect funding support for our investigation. In particular, we thank Emmo

Willinck and Lilian Parker from these organisations. The Junee Men’s Shed and Junee

Landcare assisted with the construction of the nest boxes. Darren Le Roux and two

anonymous referees provided detailed comments and background information that improved

earlier versions of this paper. Claire Shepherd assisted with manuscript preparation.

References

Barton K (2014) MuMin: multi-model inference. R-package version 1.10.5:

18

Bates D, Maechler M, Bolker B, Walker S (2014) lme4: Linear mixed-effects models using

Eigen and S4. R package: R package version 1.0-6

Beyer GL, Goldingay RL (2006) The value of nest boxes in the research and management of

Australian hollow-using arboreal marsupials. Wildlife Research 33:161-174

Bullock JM, Aronson J, Newton AC, Pywell RF, Benayas JMR (2011) Restoration of

ecosystem services and biodiversity: conflicts and opportunities. Trends in Ecology and

Evolution 26:541-549

Catterall CP, Freeman AN, Kanowski J, Freebody K (2012) Can active restoration of tropical

rainforest rescue biodiversity? A case with bird community indicators. Biological

Conservation 146:53-61

Clewell AF, Aronson J (2007) Ecological Restoration: Principles, Values, and Structure of an

Emerging Profession. Island Press, Washington, D.C.

Cooper CB, Walters JR, Ford H (2002) Effects of remnant size and connectivity on the

response of Brown Treecreepers to habitat fragmentation. Emu 102:249-256

Crane M, Cunningham RB, Lindenmayer DB (2014) The value of countryside elements in

the conservation of a threatened arboreal marsupial Petaurus norfolcensis in agricultural

landscapes of south-eastern Australia – the disproportional value of scattered trees. PLOS

One 9:e107178

Crane MJ, Lindenmayer DB, Cunningham RB (2012) Use and characteristics of nocturnal

habitats of the squirrel glider (Petaurus norfocensis) in Australian temperate woodlands.

Australian Journal of Zoology 60:320-329

Cristescu RH, Frere C, Banks PB (2012) A review of fauna in mine rehabilitation in

Australia: current state and future directions. Biological Conservation 149:60-72

19

Cunningham RB, Lindenmayer DB, Crane M, Michael D, MacGregor C (2007) Reptile and

arboreal marsupial response to replanted vegetation in agricultural landscapes. Ecological

Applications 17:609-619

Doerr VA, Doerr ED, Davies MJ (2010) Does structural connectivity facilitate dispersal of

native species in Australia's fragmented terrestrial landscapes? Available at;

http://www.environmentalevidence.org/completed-reviews/page/4: Collaboration for

Environmental Evidence. Report no. 08-007 (SR44)

Durant R, Luck GW, Matthews A (2009) Nest-box use by arboreal mammals in a peri-urban

landscape. Wildlife Research 36:565–573

Fargallo JA, Blanco G, Potti J, Vinuela J (2001) Nestbox provisioning in a rural population of

Eurasian kestrels: breeding performance, nest predation and parasitism. Bird Study 48:236-

244

Finch DM (1989) Relationships of surrounding riparian habitat to nest-box use and

reprouctive outcome in house wrens. The Condor 91:848-859

Fischer J, Stott J, Law BS (2010) The disproportionate value of scattered trees. Biological


Flaquer C, Torre I, Ruiz-Jarillo R (2006) The value of bat-boxes in the conservation of

Pipistrellus pygmaeus in wetland rice paddies. Biological Conservation 128:223-230

Gibbons P, Boak M (2002) The value of paddock trees for regional conservation in an

agricultural landscape. Ecological Management & Restoration 3:205-210

Gibbons P, Briggs SV, Murphy DY, Lindenmayer DB, McElhinny C, Brookhouse M (2010)

Benchmark stem densities for forests and woodlands in south-eastern Australia under

conditions of relatively little modification by humans since European settlement. Forest

Ecology and Management 260:2125-2133

20

Gibbons P, Lindenmayer DB (2002) Tree Hollows and Wildlife Conservation in Australia.

CSIRO Publishing, Melbourne

Gibbons P, Lindenmayer DB, Fischer J, Manning AD, Weinberg A, Sedden J, Ryan P,

Barrett G (2008) The future of scattered trees in agricultural landscapes. Conservation

Biology 22:1309-1319

Goldingay RL, Carthew SM, Whelan RJ (1991) The importance of non-flying mammals in

pollination. Oikos 61:79-87

Goldingay RL, Harrisson KA, Taylor AC, Ball TM, Sharpe DJ, Taylor BD (2013) Fine-scale

genetic response to landscape change in a gliding mammal. PLOS One 8:e880383

Goldingay RL, Rueegger NN, Grimson MJ, Taylor BD (2015) Specific nest box designs can

improve habitat restoration for cavity-dependent arboreal mammals. Restoration Ecology

23:482-490

Goldingay RL, Stevens JR (2009) Use of artificial tree hollows by Australian birds and bats.

Wildlife Research 36:81-97

Hadfield JD (2010) MCMC methods for multi-response Generalised Linear Mixed Models:

The MCMCglmm R Package. Journal of Statistical Software 33:1-22

Hajkowicz S (2009) The evolution of Australia's natural resource management programs:

Towards improved targeting and evaluation of investments. Land Use Policy 26:471-478

Karp D, Rominger AJ, Zook J, Ranganathan J, Ehrlich PR, Daily GC (2012) Intensive

agriculture erodes β−diversity at large scales. Ecology Letters 15:963-970

Lamb D (2011) Regreening the Bare Hills: Tropical Forest Regeneration in the Asia-Pacific

Region. Springer, Dordrecht, The Netherlands

Latawiec AE, Strassburg BBN, Brancalion PHS, Rodrigues RR, Gardner T (2015) Creating

space for large-scale restoration in tropical agricultural landscapes. Frontiers in Ecology and

Environment 13:211-218

21

Lindenmayer DB (2002) Gliders of Australia. A Natural History. UNSW Press, Sydney

Lindenmayer DB, Fischer J (2006) Habitat Fragmentation and Landscape Change. Island

Press, Washington, D.C.

Lindenmayer DB, Northrop-Mackie AR, Montague-Drake R, Crane M, Michael D, Okada S,

Gibbons P (2012) Not all kinds of revegetation are created equal: Revegetation type

influences bird assemblages in threatened Australian woodland ecosystems. PLOS One

7:e34527

Lindenmayer DB, Welsh A, Donnelly CF, Crane M, Michael D, MacGregor C, McBurney L,

Montague-Drake RM, Gibbons P (2009) Are nest boxes a viable alternative source of cavities

for hollow-dependent animals? Long-term monitoring of nest box occupancy, pest use and

attrition. Biological Conservation 142:33-42

Loos J, Abson DJ, Chappell MJ, Hanspach J, Mikulcak F, Tichit M, Fischer J (2014) Putting

meaning back into "sustainable intensification". Frontiers in Ecology and Environment

12:356-361

Manning AD, Fischer J, Lindenmayer DB (2006) Scattered trees are keystone structures -

Implications for conservation. Biological Conservation 132:311-321

Menkhorst P (1984) The application of nest boxes in research and management of possums

and gliders. Pages 517-525 In: Smith AP, Hume ID, eds) Possums and Gliders. Surrey Beatty

and Sons, Melbourne

Menz M, Dixon K, Hobbs R (2013) Hurdles and opportunities for landscape-scale

restoration. Science 339:526-527

Minnemeyer S, Laestadius L, Sizer N, Saint-Laurent C, Potapov P (2011) A World of

Opportunity. Washington, D.C. Available at:

http://www.wri.org/sites/default/files/world_of_opportunity_brochure_2011-09.pdf. World

Resources Institute

22

Munro N, Lindenmayer DB (2011) Planting for Wildlife: A Practical Guide to Restoring

Native Woodlands. CSIRO Publishing, Melbourne

Pell AS, Tidemann CR (1997) The impact of two exotic hollow-nesting birds on two native

parrots in svannah and woodland in eastern Australia. Biological Conservation 79:145-153

Ray Benayas JM, Newton AC, Diaz A, Bullock JM (2009) Enhancement of biodiversity and

ecosystem services by ecological restoration: a meta-analysis. Science 325:1112-1124

Smith GC, Agnew G (2002) The value of "bat boxes" for attracting hollow-dependent fauna

to farm forestry plantations in southeast Queensland. Ecological Management & Restoration

3:37-46

Soderquist TR, MacNally R (2000) The conservation value of mesic gullies in dry forest

landscapes: mammal populations in the box-ironbark ecosystem of southern Australia.

Biological Conservation 93:281-291

Suding K, et al. (2015) Committing to ecological restoration. Science 348:638-640

van der Ree R, Bennett AF (2003) Home range of the squirrel glider (Petaurus norfolcensis)

in a network of remnant linear habitats. Journal of Zoology 259:327-336

van der Ree R, Bennett AF, Gilmore DC (2004) Gap-crossing by gliding marsupials:

thresholds for use of isolated woodland patches in an agricultural landscape. Biological


Vesk P, Nolan R, Thomson JW, Dorrough JW, Mac Nally R (2008) Time lags in the

provision of habitat resources through revegetation. Biological Conservation 141:174-186

Wood S (2006) Generalized Additive Models, An Introduction with R. Chapman and Hall,

London

Wortley L, Hero J-M, Howes M (2013) Evaluating ecological restoration sucecss: a review of

the literature. Restoration Ecology 21:537-543

23

Table 1. Summary data showing the percentage occupancy of nest boxes by different species. Values in brackets are numbers of

occupied boxes. Abbreviations are as follows: Tvu (Trichosurus vulpecula; Common Brushtail Possum); Rra (Rattus rattus, Black

Rat); Ppe (Pseudocheirus peregrinus; Common Ringtail Possum); Pbr (Petaurus breviceps; Sugar Glider); Afl (Antechinus flavipes;

Yellow-footed Antechinus); Nge (Nyctophilus geoffroyii; Lesser Long-eared Bat); Svu (Sturnus vulgaris; Common Starling); Pex

(Platycercus eximus; Eastern Rosella); Ero (Eolophus eximius; Galah); Nho (Nymphicus hollandicus; Cockatiel); Lper (Litoria

peronii; Peron’s Tree Frog); Cma (Christinus marmoratus; Marbled Gecko) and Ame (Apis mellifera; Feral Honeybee). Exotic

species are denoted by a star (*)

Percentage of boxes used

Nest box type Tvu Rra* Ppe Pbr Afl Nge Svu* Pex Ero Nho Lper Cma Ame* No

evidence

of use

Brushtail Possum

box (44)

36%

(16)

9% (4) 9% (4) 2% (1) 2% (1) 0 43%

(19)

11% (5) 0 0 7% (3) 2% (1) 31%

(14)

2% (1)

Kookaburra box

(16)

38% (6) 13% (2) 6% (1) 0 0 0 13% (2) 0 0 0 0 19% (3) 13% (2)

Squirrel Glider

box (60)

5% (3) 18%

(11)

2% (1) 2% (1) 7% (4) 5% (3) 65%

(39)

8% (5) 2% (1) 2% (1) 5% (3) 2% (1) 33%

(20)

3% (2)

Superb Parrot box

(30)

33%

(10)

13% (4) 7% (2) 0 0 0 43%

(13)

23% (6) 0 0 0 0 40%

(12)

0

24

Figure 1. The location of the study area and field sites (16 plantings and 14 remnants) where 6

nest boxes were established and checked four times between 2010 and 2012. 7

8

Figure 2. Nest box occupancy for species groups and individual species (with associated ±95% 9

credible intervals). Black credible intervals correspond to the final models constructed from the 10

broad scale variables only, whereas, the blue intervals correspond to the final model after 11

inclusion of the fine scale variables. Absence of credible intervals indicate that a given variable 12

vegetation was not important in the broad scale analysis. A. Nest box occupancy in relation to 13

broad vegetation type (remnants versus plantings). The y-axis is on the log-odds scale: log (odds 14

of Remnant/ odds Planting), values greater than 0 indicate a preference for remnant patches, 15

whereas values less than 0 indicate a preference for plantings. B. Nest box occupancy in relation 16

to connected and unconnected remnants and plantings. Absence of credible intervals indicate that 17

connectivity was not important in the broad scale analysis. The y-axis is on the log-odds scale: 18

log (odds of Isolated/ odds of Connected), values greater than 0 indicate a preference for isolated 19

patches, whereas values less than 0 indicate a preference for connected patches. C. Nest box 20

occupancy in relation to the distance to drainage line. An absence of credible intervals indicate 21

that distance to drainage line was not important in the fine scale analysis. The y-axis corresponds 22

to the linear slope of distance to drainage lines, values less than 0 indicate a negative association 23

between distance from drainage line and presence of the indicated species or species group. 24

Abbreviations are as follows: CBP (Common Brushtail Possum), CS (Common Starling), FHB 25

(Feral Honeybee), Any (any species). 26

27

28

25

Figure 1 29

30

Figure 2 31

32

1

SUPPLEMNTARY TABLE AND APPENDICES

Table S1: Dimensions of the four different types of nest boxes deployed in this study

Box type Height

(mm)

Depth

(mm)

Width

(mm)

Hole size

(mm)

Bat

chamber

Common Brushtail

Possum Box

500 350 360 80 no

Superb Parrot Box 550 250 260 80 no

Squirrel Glider Box 500 230 240 45 yes

Kookaburra Box 250 260 550 90 yes

2

Appendix 1: Broad scale variables

Appendix 1A: Presence of any mammal species – Posterior summary of Random effects

logistic regression model. Abbreviations Sm and S correspond to summer and spring,

respectively.

Fixed Effects:

Parameter Posterior mean l-95% CI u-95% CI Btail

Intercept (SurveyOcc: 2010.S,

Connectivity: Connected)

-3.199 -2.233 -4.148 <0.001

SurveyOcc 2011.S 2.133 3.014 1.219 <0.001

SurveyOcc 2011.Sm 1.073 1.984 0.055 0.012

SurveyOcc 2012.S 1.630 2.519 0.719 <0.001

Connectivity Isolated -0.937 0.007 -1.925 0.029

Additional Comparisons

SurveyOcc: 2011.S vs 2011.Sm 1.060 0.332 1.748 0.002

SurveyOcc: 2011.S vs 2012.S 0.503 -0.154 1.163 0.068

SurveyOcc: 2011.Sm vs 2012.S -0.557 -1.317 0.212 0.078

Random Effects:

Parameter Posterior mean l-95% CI u-95% CI

SiteCode RE 1.221 0.452 2.144

Observation Level RE 0.574 0.184 1.121

Appendix 1B: Presence of Common Brushtail Possum – Posterior summary of Random

effects logistic regression model. Abbreviations Sm and S correspond to summer and spring,

respectively.

3

Fixed Effects:



NestBoxType: BP, Connectivity:

Connected)

-3.104 -1.982 -4.172 <0.001

SurveyOcc 2011.S 1.241 2.258 0.293 0.008

SurveyOcc 2011.Sm 0.675 1.576 -0.394 0.098

SurveyOcc 2012.S 0.449 1.54 -0.577 0.188

Connectivity Isolated -2.357 -0.877 -3.829 <0.001


SurveyOcc: 2011.S vs 2011.Sm 0.566 -0.348 1.48 0.113

SurveyOcc: 2011.S vs 2012.S 0.792 -0.144 1.71 0.052

SurveyOcc: 2011.Sm vs 2012.S 0.226 -0.732 1.173 0.318

Random Effects:


SiteCode RE 1.794 0.586 3.257


Appendix 1C: Presence of any bird species – Posterior summary of Random effects logistic

regression model. Abbreviations Sm and S correspond to summer and spring, respectively.

Fixed Effects:


Intercept (SurveyOcc: 2010.S) -3.870 -2.868 -4.891 <0.001

SurveyOcc 2011.S 2.575 3.437 1.629 <0.001

4

SurveyOcc 2011.Sm 2.785 3.716 1.872 <0.001

SurveyOcc 2012.S 2.957 3.913 2.079 <0.001


SurveyOcc: 2011.S vs 2011.Sm -0.209 -0.871 0.444 0.268

SurveyOcc: 2011.S vs 2012.S -0.381 -0.998 0.275 0.117

SurveyOcc: 2011.Sm vs 2012.S -0.172 -0.811 0.433 0.293

Random Effects:


SiteCode RE 2.018 0.815 3.487


Appendix 1D: Presence of Common Starling – Posterior summary of Random effects


respectively.

Fixed Effects:


Intercept (SurveyOcc: 2010.S) -4.018 -2.986 -5.093 <0.001

SurveyOcc 2011.S 2.547 3.577 1.594 <0.001

SurveyOcc 2011.Sm 2.327 3.448 1.411 <0.001

SurveyOcc 2012.S 2.859 3.816 1.791 <0.001


SurveyOcc: 2011.S vs 2011.Sm 0.220 -0.438 0.903 0.267

SurveyOcc: 2011.S vs 2012.S -0.312 -0.965 0.374 0.174

SurveyOcc: 2011.Sm vs 2012.S -0.532 -1.172 0.115 0.056

5

Random Effects:


SiteCode RE 1.871 0.768 3.181


Appendix 1E: Presence of Feral Honeybees – Posterior summary of Random effects logistic


Fixed Effects:



VegetationType: Planting)

-3.743 -1.636 -6.021 <0.001

SurveyOcc 2011.S 2.734 4.273 1.215 <0.001

SurveyOcc 2011.Sm 2.642 4.127 1.213 <0.001

SurveyOcc 2012.S 4.399 5.948 2.940 <0.001

Connectivity Isolated 0.702 1.731 -0.333 0.086

VegetationType Remnant -2.168 -1.103 -3.269 <0.001

No. of paddock trees in 500m -0.990 0.032 -2.132 0.030


SurveyOcc: 2011.S vs 2011.Sm 0.092 -0.818 0.931 0.416

SurveyOcc: 2011.S vs 2012.S -1.665 -2.483 -0.894 <0.001

SurveyOcc: 2011.Sm vs 2012.S -1.757 -2.608 -1.012 <0.001

6

Random Effects:


SiteCode RE 0.847 0.283 1.557


Appendix 1F: Presence of any species – Posterior summary of Random effects logistic


Fixed Effects:



Vegetation Type: Planting)

-1.935 -0.836 -2.916 <0.001

SurveyOcc 2011.S 3.225 4.038 2.436 <0.001

SurveyOcc 2011.Sm 2.748 3.568 2.080 <0.001

SurveyOcc 2012.S 4.736 5.666 3.781 <0.001

VegetationType Remnant -1.407 -0.091 -2.639 0.014


SurveyOcc: 2011.S vs 2011.Sm 0.477 -0.118 1.098 0.066

SurveyOcc: 2011.S vs 2012.S -1.511 -2.345 -0.726 <0.001

SurveyOcc: 2011.Sm vs 2012.S -1.988 -2.793 -1.25 <0.001

Random Effects:


SiteCode RE 2.920 1.154 4.930


7

Appendix 2. Broad + fine scale variables

(Broad scale variables remain in the model and the model selection is done on which fine

scale variables are important)

Appendix 2A: Presence of any mammal species – Posterior summary of Random effects


respectively.

Fixed Effects:




Connected)

-3.356 -2.262 -4.584 <0.001

SurveyOcc 2011.S 2.374 3.382 1.348 <0.001

SurveyOcc 2011.Sm 1.171 2.200 0.117 0.010

SurveyOcc 2012.S 1.779 2.821 0.890 <0.001

Connectivity Isolated -0.922 0.035 -2.065 0.036

NestBoxType KB -0.045 0.794 -0.964 0.475

NestBoxType SG -1.426 -0.732 -2.085 <0.001

Dieback Score 0.320 0.636 -0.029 0.026

Distance to drainage line -0.828 -0.354 -1.349 <0.001

Tree Diameter -0.506 -0.100 -0.915 0.009


SurveyOcc: 2011.S vs 2011.Sm 1.203 0.387 2.015 0.001

SurveyOcc: 2011.S vs 2012.S 0.595 -0.109 1.301 0.052

SurveyOcc: 2011.Sm vs 2012.S -0.607 -1.489 0.194 0.072

NestBoxType: KB vs SG 1.381 0.434 2.335 0.004

8

Random Effects:


SiteCode RE 1.162 0.404 2.180


Appendix 2B: Presence of Common Brushtail Possum – Posterior summary of Random

effects logistic regression model. Abbreviations Sm and S correspond to summer and spring,

respectively.

Fixed Effects:




Connected)

-4.619 -2.703 -6.565 <0.001

SurveyOcc 2011.S 1.710 2.961 0.609 0.002

SurveyOcc 2011.Sm 0.911 2.110 -0.244 0.068

SurveyOcc 2012.S 0.492 1.775 -0.616 0.196

Connectivity Isolated -2.665 -1.024 -4.811 0.002

NestBoxType KB 0.438 1.477 -0.666 0.212

NestBoxType SG -3.557 -2.189 -5.091 <0.001

Dieback Score 0.565 1.034 0.005 0.014

Distance to drainage line -1.224 -0.347 -2.044 0.002

logStemsP1 0.928 1.886 -0.029 0.028


SurveyOcc: 2011.S vs 2011.Sm 0.799 -0.304 1.921 0.077

SurveyOcc: 2011.S vs 2012.S 1.218 0.130 2.321 0.018

9

SurveyOcc: 2011.Sm vs 2012.S 0.419 -0.713 1.536 0.240

NestBoxType: KB vs SG 3.996 2.352 5.803 <0.001

Random Effects:


SiteCode RE 2.686 0.748 5.259


Appendix 2C: Presence of any bird species – Posterior summary of Random effects logistic


Fixed Effects:



NestBoxType: BP)

-4.553 -3.411 -5.735 <0.001

SurveyOcc 2011.S 2.917 3.926 1.912 <0.001

SurveyOcc 2011.Sm 3.158 4.215 2.141 <0.001

SurveyOcc 2012.S 3.379 4.429 2.408 <0.001

NestBoxType KB -3.300 -1.650 -4.831 <0.001

NestBoxType SG 0.964 1.507 0.416 <0.001

Tree Diameter 0.313 0.631 -0.072 0.039


SurveyOcc: 2011.S vs 2011.Sm -0.240 -0.906 0.446 0.248

SurveyOcc: 2011.S vs 2012.S -0.462 -1.105 0.180 0.084

SurveyOcc: 2011.Sm vs 2012.S -0.221 -0.892 0.458 0.259

NestBoxType: KB vs SG -4.263 -6.046 -2.715 <0.001

10

Random Effects:


SiteCode RE 2.744 1.162 4.891


Appendix 2D: Presence of Common Starling – Posterior summary of Random effects


respectively.

Fixed Effects:



NestBoxType: BP)

-4.813 -3.592 -5.989 <0.001

SurveyOcc 2011.S 2.823 3.831 1.796 <0.001

SurveyOcc 2011.Sm 2.584 3.678 1.577 <0.001

SurveyOcc 2012.S 3.222 4.247 2.167 <0.001

NestBoxType KB -2.587 -1.038 -4.137 <0.001

NestBoxType SG 1.226 1.777 0.625 <0.001


SurveyOcc: 2011.S vs 2011.Sm 0.239 -0.471 0.962 0.260

SurveyOcc: 2011.S vs 2012.S -0.399 -1.092 0.299 0.127

SurveyOcc: 2011.Sm vs 2012.S -0.638 -1.355 0.064 0.040


11

Random Effects:


SiteCode RE 2.424 0.933 4.129


Appendix 2E: Presence of Feral Honeybees – Posterior summary of Random effects logistic


Fixed Effects:



NestBoxType: BP,

VegetationType: Planting)

-5.776 -3.237 -8.305 <0.001

SurveyOcc 2011.S 2.966 4.785 1.379 <0.001

SurveyOcc 2011.Sm 2.831 4.617 1.264 <0.001

SurveyOcc 2012.S 4.771 6.447 3.085 <0.001

Connectivity Isolated 0.582 1.577 -0.439 0.122

VegetationType Remnant -1.109 0.224 -2.451 0.047

NestBoxType KB -1.415 -0.159 -3.021 0.018

NestBoxType SG 0.362 1.067 -0.280 0.146

No. of paddock trees in 500m -0.710 0.246 -1.699 0.066

logStemsP1 0.760 1.609 -0.051 0.034

Distance to closest major veg 0.470 1.091 -0.159 0.070


SurveyOcc: 2011.S vs 2011.Sm 0.135 -0.818 1.126 0.392

SurveyOcc: 2011.S vs 2012.S -1.806 -2.654 -0.983 <0.001

12

SurveyOcc: 2011.Sm vs 2012.S -1.941 -2.863 -1.116 <0.001

NestBoxType: KB vs SG -1.777 -3.359 -0.405 0.006

Random Effects:


SiteCode RE 0.682 0.251 1.329


Appendix 2F: Presence of any species – Posterior summary of Random effects logistic


Fixed Effects:



NestBoxType: BP, Vegetation

Type: Planting)

-2.837 -1.603 -4.006 <0.001

SurveyOcc 2011.S 3.550 4.405 2.790 <0.001

SurveyOcc 2011.Sm 3.012 3.839 2.252 <0.001

SurveyOcc 2012.S 5.290 6.290 4.274 <0.001

VegetationType Remnant 0.022 1.654 -1.667 0.494

NestBoxType KB -2.129 -1.204 -3.047 <0.001

NestBoxType SG 0.526 1.081 -0.065 0.037

Distance to drainage line -0.463 0.103 -1.027 0.054

Distance to closest major veg 1.271 2.256 0.440 0.002


SurveyOcc: 2011.S vs 2011.Sm 0.538 -0.133 1.224 0.064

13

SurveyOcc: 2011.S vs 2012.S -1.739 -2.583 -0.929 <0.001

SurveyOcc: 2011.Sm vs 2012.S -2.278 -3.174 -1.450 <0.001


Random Effects:


SiteCode RE 3.025 1.256 5.151


Do nest boxes in restored woodlands promote the ......1 Restoration Ecology Do nest boxes in restored woodlands promote the conservation of hollow-dependent fauna? David Lindenmayer1,2,3,

Documents